A new metric to assess reference populations for genomic selection in Australian beef breeds

Research has shown that to maximise benefits (i.e. EBV accuracy and spread and decreased generation interval) of genomic selection, reference populations of genotyped and phenotyped animals need to be large, diverse in relationships and closely related to the selection generation. In practice, applying these design principles can be challenging as metrics to aid decision making are often not available. This paper applies a metric that objectively describes reference populations and their impact on accuracy for seven Australian beef populations. All populations for live weight and abattoir carcase had different levels of phenotyping, genotyping and overall size. Carcase traits had the lowest reference population size, and assessment of average relatedness in almost all the breeds showed there were herds that were not currently represented in the reference population. Generally, increasing reference size resulted in increased accuracy, but there were some exceptions. The metrics described in this paper are easy to apply and can assist in the construction of reference populations.